This overview aims to stimulate conceptual and practical discussions to help unlock the full potential of realist evaluation in health systems research. Based on a structured literature search, this review maps how the concepts of realist evaluation are applied in health systems research and which methodological problems are encountered. We found a great diversity in the depth of application of the philosophical concepts, use of terminology and scope of application in the research process. Terms of theory-driven evaluation, theories of change and realist evaluation are often used interchangeably. Diverging views exist regarding the nature of ‘mechanism’ and the difference between mechanism and essential context condition. A lack of methodological guidance was highlighted by many authors. Realist evaluation is slowly gaining traction in health systems research, but more clarity is needed concerning the definitions of mechanisms and context and how the configuration of context, mechanism and intervention can be described and assessed.

Frameworks can clarify concepts and improve understanding of underlying mechanisms in the domain of health systems research and strengthening. Many existing frameworks have a limited capacity to analyze interactions and equilibriums within a health system overlooking values as an underlying steering mechanism. This paper introduces the health system dynamics framework and demonstrates its application as a tool for analysis and modelling. The added value of this framework is: 1) consideration of different levels of a health system and tracing how interventions or events at one level influence other elements and other levels; 2) emphasizes the importance of values; 3) a central axis linking governance, human resources, service delivery and population, and 4) taking into account the key elements of complexity in analysis and strategy development. We urge the analysis of individual health systems and meta-analysis, for a better understanding of their functioning and strengthening.

Despite the mounting attention for health systems and health systems theories, there is a persisting lack of consensus on their conceptualisation and strengthening. This paper contributes to structuring the debate, presenting landmarks in the development of health systems thinking against the backdrop of the policy context and its dominant actors. We argue that frameworks on health systems are products of their time, emerging from specific discourses. They are purposive, not neutrally descriptive, and are shaped by the agendas of their authors. The evolution of thinking over time does not reflect a progressive accumulation of insights. Instead, theories and frameworks seem to develop in reaction to one another, partly in line with prevailing paradigms and partly as a response to the very different needs of their developers. The reform perspective considering health systems as projects to be engineered is fundamentally different from the organic view that considers a health system as a mirror of society. The co-existence of health systems and disease-focused approaches indicates that different frameworks are complementary but not synthetic. The contestation of theories and methods for health systems relates almost exclusively to low income countries. At the global level, health system strengthening is largely narrowed down to its instrumental dimension, whereby well-targeted and specific interventions are supposed to strengthen health services and systems or, more selectively, specific core functions essential to programmes. This is in contrast to a broader conceptualization of health systems as social institutions. Health systems theories and frameworks frame health, health systems and policies in particular political and public health paradigms. While there is a clear trend to try to understand the complexity of and dynamic relationships between elements of health systems, there is also a demand to provide frameworks that distinguish between health system interventions, and that allow mapping with a view of analysing their returns. The choice for a particular health system model to guide discussions and work should fit the purpose. The understanding of the underlying rationale of a chosen model facilitates an open dialogue about purpose and strategy.

Background: The increasing prevalence of chronic diseases puts a high burden on the health care systems of Low and Middle Income Countries which are often not adapted to provide the care needed. Peer support programmes are promoted to address health system constraints. This case study analyses a peer educator diabetes programme in Cambodia, MoPoTsyo, from a health system’s perspective. Which strategies were used and how did these strategies change? How is the programme perceived? Methods: Data were collected through semi-structured interviews with patients, MoPoTsyo staff and peer educators, contracted pharmacy staff and health workers, health care workers and non-contracted pharmacists and managers and policy makers at district, provincial and national level. Four areas were purposively selected to do the interviews. An inductive content analysis was done independently by two researchers. Results: MoPoTsyo developed into three stages: a focus on diabetes self-management; a widening scope to ensure affordable medicines and access to other health care services; and aiming for sustainability through more integration with the Cambodian public system and further upscaling. All respondents acknowledged the peer educators’ role and competence in patient education, but their ideas about additional tasks and their place in the system differed. Indirectly involved stakeholders and district managers emphasized the particular roles and responsibilities of all actors in the system and the particular role of the peer educator in the community. MoPoTsyo’s diagnostics and laboratory services were perceived as useful, especially by patients and project staff. Respondents were positive about the revolving drug fund, but expressed concerns about its integration into the government system. The degree of collaboration between health care staff and peer educators varied. Conclusion: MoPoTsyo responds to the needs of people with diabetes in Cambodia. Key success factors were: consistent focus on and involvement of the target group, backed up by a strong organisation; simultaneous reduction of other barriers to care; and the ongoing maintenance of relations at all levels within the health system. Despite resistance, MoPoTsyo has established a more balanced relationship between patients and health service providers, empowering patients to self-manage and access services that meet their needs.

The growing caseload caused by patients with chronic life-long conditions leads to increased needs for health care providers and rising costs of health services, resulting in a heavy burden on health systems, populations and individuals. The professionalised health care for chronic patients common in high income countries is very labour-intensive and expensive. Moreover, the outcomes are often poor. In low-income countries, the scarce resources and the lack of quality and continuity of health care result in high health care expenditure and very poor health outcomes. The current proposals to improve care for chronic patients in low-income countries are still very much provider-centred.

Process evaluation is increasingly recognized as an important component of effective implementation research and yet, there has been surprisingly little work to understand what constitutes best practice. Researchers use different methodologies describing causal pathways and understanding barriers and facilitators to implementation of interventions in diverse contexts and settings. We report on challenges and lessons learned from undertaking process evaluation of seven hypertension intervention trials funded through the Global Alliance of Chronic Diseases (GACD). Preliminary data collected from the GACD hypertension teams in 2015 were used to inform a template for data collection. Case study themes included: (1) description of the intervention, (2) objectives of the process evaluation, (3) methods including theoretical basis, (4) main findings of the study and the process evaluation, (5) implications for the project, policy and research practice and (6) lessons for future process evaluations. The information was summarized and reported descriptively and narratively and key lessons were identified. The case studies were from low- and middle-income countries and Indigenous communities in Canada. They were implementation research projects with intervention arm. Six theoretical approaches were used but most comprised of mixed-methods approaches. Each of the process evaluations generated findings on whether interventions were implemented with fidelity, the extent of capacity building, contextual factors and the extent to which relationships between researchers and community impacted on intervention implementation. The most important learning was that although process evaluation is time consuming, it enhances understanding of factors affecting implementation of complex interventions. The research highlighted the need to initiate process evaluations early on in the project, to help guide design of the intervention; and the importance of effective communication between researchers responsible for trial implementation, process evaluation and outcome evaluation. This research demonstrates the important role of process evaluation in understanding implementation process of complex interventions. This can help to highlight a broad range of system requirements such as new policies and capacity building to support implementation. Process evaluation is crucial in understanding contextual factors that may impact intervention implementation which is important in considering whether or not the intervention can be translated to other contexts.